VerifiedEvidence: highv1.0.0

Semi-Parametric Survival

A category of survival models, notably the Cox model, that assumes covariate effects parametrically while leaving the baseline hazard unspecified.

Last reviewedDarrin Baines IP Ltd

Concept Architecture

Concept


Theoretically, Semi-Parametric Survival refers to survival modelling approaches that combine parametric and non-parametric components, allowing estimation of covariate effects without requiring specification of the underlying baseline hazard function. The Cox proportional hazards model is the canonical semi-parametric survival model. Semi-parametric survival methods exist because they provide flexibility in modelling time-to-event data while retaining efficient estimation of treatment effects and prognostic factors.

Mathematically, semi-parametric survival models represent the hazard function as the product of an unspecified baseline hazard and a parametric component describing the effects of explanatory variables. Regression coefficients are estimated using partial likelihood rather than the full likelihood, eliminating the need to estimate the baseline hazard directly. This framework provides hazard ratio estimates while leaving the baseline hazard unconstrained.

In practice, semi-parametric survival models are widely applied in clinical trials, observational studies and health technology assessments to estimate treatment effects on survival. They are routinely used in health economic evaluation to inform decision-analytic models, compare interventions and estimate relative treatment effectiveness before survival extrapolation.


Purpose


Used to estimate the effects of covariates on survival outcomes without specifying the functional form of the baseline hazard, providing flexible and robust survival modelling for health economic evaluation.


Mathematical Formulae

Primary Formula

h(t | X) = h?(t) ? exp(??X)

Supporting Formulae

Hazard ratio:

HR = exp(?)

Partial likelihood:

L(?) = ? exp(??x?) / ???R(t?) exp(??x?)

Survival function:

S(t | X) = S?(t)^exp(??X)

Related Mathematical Methods

  • Cox Proportional Hazards Model
  • Partial Likelihood Estimation
  • Kaplan?Meier Estimation
  • Schoenfeld Residuals
  • Proportional Hazards Assumption
  • Survival Analysis

Example


A health technology assessment compares a new cancer treatment with standard care using a Cox proportional hazards model.

Estimated regression coefficient:

?? = ?0.511

Hazard ratio:

HR = exp(?0.511) = 0.60

The model estimates a 40% reduction in the hazard of death for the new treatment without requiring specification of the baseline hazard function.


Excel Implementation

FunctionExample FormulaHealth Economics Application
EXP=EXP(B2)Convert regression coefficients into hazard ratios
LN=LN(C2)Calculate log hazard ratios
IF=IF(D2<0.05,"Significant","Not Significant")Interpret regression results
PRODUCT=PRODUCT(E2:E20)Illustrate calculation of partial likelihood components
EXP=EXP(-H0*EXP(B2)*A2)Calculate survival probabilities using an estimated baseline hazard

VBA (Optional)


A VBA procedure can automate estimation summaries, hazard ratio calculations and diagnostic reporting for semi-parametric survival models exported from statistical software.


Sources

  • Cox DR. Regression Models and Life-Tables. Journal of the Royal Statistical Society: Series B. 1972.
  • Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model.
  • Collett D. Modelling Survival Data in Medical Research.
  • Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
  • NICE. Health Technology Evaluation Manual.

Library

Tools & Resources

1
  • Other

    survHE — Survival Analysis for Health Economic Evaluation (R package) — Gianluca Baio, R package ed., 2023 (CRAN)

    An R package for fitting and comparing parametric survival models for health economic evaluation, including Bayesian estimation, and for extrapolating time-to-event data to inform cost-effectiveness models.

Frequently Asked Questions (6)

  • What is semi-parametric survival?

    A category of survival models, notably the Cox model, that assumes covariate effects parametrically while leaving the baseline hazard unspecified.

    Source: Cox 1972

  • What is left unspecified in a semi-parametric survival model?

    A semi-parametric survival model specifies part of its structure with parameters and leaves part unspecified. In the Cox model the way covariates such as treatment shift the hazard is captured by estimated parameters, while the underlying baseline hazard, how risk changes over time absent any covariate, is left free to take any shape and is not modelled. This lets covariate effects be estimated without committing to a form for the baseline. Leaving the baseline unspecified is what makes it only semi-parametric. Collett (2015) explains this.

    Source: Collett 2015

  • Why is the Cox model called semi-parametric?

    The Cox model is called semi-parametric because it has both parametric and non-parametric parts: it models the effect of covariates on the hazard parametrically, through the exponential of a linear combination of covariates giving hazard ratios, while leaving the baseline hazard, the hazard's shape over time, completely unspecified. This mixture, parametric covariate effects and a non-parametric baseline, is the defining feature of a semi-parametric model, distinguishing it from fully parametric models that assume a distribution and non-parametric methods that assume nothing.

    Source: Cox 1972

  • What are the advantages of semi-parametric survival models?

    Semi-parametric survival models, such as the Cox model, avoid assuming a particular distribution for the baseline hazard, so they are robust to its shape and applicable in many settings where the hazard's form is unknown, while still estimating interpretable covariate effects as hazard ratios. They handle censoring and multiple covariates and are widely applicable. This flexibility, not requiring a distributional assumption for the baseline yet quantifying covariate effects, is why the Cox model is the most common regression approach in survival analysis.

    Source: Kalbfleisch & Prentice 2002

  • What are the limitations of semi-parametric survival models?

    Semi-parametric models, by leaving the baseline hazard unspecified, do not provide a full parametric survival curve, so they are less suited to extrapolation beyond the data, for which a fully parametric model is needed. The Cox model also relies on the proportional hazards assumption, which must be checked. Estimating absolute survival requires additional steps to recover the baseline hazard. These limitations mean semi-parametric models are excellent for estimating covariate effects but complemented by parametric models where absolute survival or extrapolation is required.

    Source: Cox 1972

  • How does semi-parametric survival differ from parametric survival?

    Semi-parametric survival models specify covariate effects parametrically but leave the baseline hazard unspecified, while fully parametric survival models assume a specific distribution for the entire hazard or survival. The semi-parametric approach is robust to the baseline shape and focuses on covariate effects, but it does not readily give a full survival curve or extrapolation, whereas parametric models provide smooth, extendable survival curves at the cost of assuming a distribution. The two are complementary, with semi-parametric models for covariate effects and parametric ones for absolute survival and extrapolation.

    Source: Collett 2015

Trust Record

Verified by Dr Darrin Baines

British health economist

Professional identity: darrinbaines.org

Verification date: 23 Oct 2025

Content version: 1.0.0

Canonical Identity

Term code
HE-EM-SM-078

Stable URI · Machine-readable · Resolvable · CC BY 4.0